QString tidyHtml(QString str, bool& ok) {
#ifdef NO_TIDY
ok = true;
return str;
#else
QString res = str;
ok = false;
static bool isTidyWithIntBodyOnly = isTidyWithIntBodyOnlyCheck();
TidyDoc tdoc = tidyCreate();
TidyBuffer output;
TidyBuffer errbuf;
tidyBufInit(&output);
tidyBufInit(&errbuf);
bool configOk =
tidyOptSetBool(tdoc, TidyXhtmlOut, yes) &&
tidyOptSetBool(tdoc, TidyForceOutput, yes) &&
tidyOptSetBool(tdoc, TidyMark, no) &&
(isTidyWithIntBodyOnly
? tidyOptSetInt(tdoc, TidyBodyOnly, 1)
: tidyOptSetBool(tdoc, TidyBodyOnly, yes)) &&
tidyOptSetInt(tdoc, TidyWrapLen, 0) &&
tidyOptSetInt(tdoc, TidyDoctypeMode, TidyDoctypeOmit);
if (configOk &&
(tidySetCharEncoding(tdoc, "utf8") >= 0) &&
(tidySetErrorBuffer(tdoc, &errbuf) >= 0) &&
(tidyParseString(tdoc, str.toUtf8().data()) >= 0) &&
(tidyCleanAndRepair(tdoc) >= 0) &&
(tidyRunDiagnostics(tdoc) >= 0) &&
(tidySaveBuffer(tdoc, &output) >= 0) &&
(output.bp != 0 && output.size > 0)) {
res = QString::fromUtf8((char*)output.bp, output.size);
ok = true;
}
#ifdef DEBUG_MARKUP
if (errbuf.size > 0) {
QString errStr = QString::fromUtf8((char*)errbuf.bp, errbuf.size);
qDebug() << "\n[DEBUG] MARKUP, libtidy errors and warnings:\n" << errStr;
}
#endif
if (output.bp != 0)
tidyBufFree(&output);
if (errbuf.bp != 0)
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return res.trimmed();
#endif
}
QVector<QString> UCHome_Main_SiteConst::find_sys_notice(TidyDoc doc)
{
QVector<QString> sys_notes;
QString note;
char *notes_ids[] = { "mtag_invite", "f_request", NULL };
char *nid = NULL;
TidyNode node = NULL;
TidyNode node2 = NULL;
TidyBuffer tbuf;
for(int i = 0; i < 16 ; i++) {
nid = notes_ids[i];
if(nid == NULL) {
break;
}
node = this->searchNode(doc, NULL, nid, TidyTag_DIV);
if(node != NULL) {
tidyBufInit(&tbuf);
tidyNodeGetText(doc, node, &tbuf);
note = this->u8codec->toUnicode(QByteArray((char*)tbuf.bp));
sys_notes.append(note);
tidyBufFree(&tbuf);
q_debug()<<note;
}else{
q_debug()<<"Warning: "<<nid<<" not found";
}
}
return sys_notes;
}
void CCaHtmlParse::__GetPriceAndRamainTicket(UINT *pPrice, UINT *pRemainTicket, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
*pPrice = 0;
*pRemainTicket = 0;
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
CStringA straKey("</font>");
int iPos = straRet.Find(straKey);
int iEndPos = straRet.Find("</strong>");
int iStartPos = iPos+straKey.GetLength();
CStringA straPrice = straRet.Mid(iStartPos, iEndPos-iStartPos);
straPrice.Remove(0x0d);//去掉回车
straPrice.Remove(0x0a);//去掉换行
*pPrice = atoi(straPrice.GetBuffer(0));
straPrice.ReleaseBuffer();
//剩余座位
straKey = "：";
iPos = straRet.Find(straKey);
iEndPos = straRet.Find("</td>");
iStartPos = iPos+straKey.GetLength();
CStringA straRemainSeat = straRet.Mid(iStartPos, iEndPos-iStartPos);
*pRemainTicket = atoi(straRemainSeat.GetBuffer(0));
straRemainSeat.ReleaseBuffer();
tidyBufFree(&text);
}
static int tidy_doc_cast_handler(zval *in, zval *out, int type)
{
TidyBuffer output;
PHPTidyObj *obj;
switch (type) {
case IS_LONG:
ZVAL_LONG(out, 0);
break;
case IS_DOUBLE:
ZVAL_DOUBLE(out, 0);
break;
case _IS_BOOL:
ZVAL_TRUE(out);
break;
case IS_STRING:
obj = Z_TIDY_P(in);
tidyBufInit(&output);
tidySaveBuffer (obj->ptdoc->doc, &output);
ZVAL_STRINGL(out, (char *) output.bp, output.size ? output.size-1 : 0);
tidyBufFree(&output);
break;
default:
return FAILURE;
}
return SUCCESS;
}
static void tidy_doc_update_properties(PHPTidyObj *obj)
{
TidyBuffer output;
zval temp;
tidyBufInit(&output);
tidySaveBuffer (obj->ptdoc->doc, &output);
if (output.size) {
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
ZVAL_STRINGL(&temp, (char*)output.bp, output.size-1);
zend_hash_str_update(obj->std.properties, "value", sizeof("value") - 1, &temp);
}
tidyBufFree(&output);
if (obj->ptdoc->errbuf->size) {
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
ZVAL_STRINGL(&temp, (char*)obj->ptdoc->errbuf->bp, obj->ptdoc->errbuf->size-1);
zend_hash_str_update(obj->std.properties, "errorBuffer", sizeof("errorBuffer") - 1, &temp);
}
}
/* Traverse the document tree */
void dumpNode(TidyDoc doc, TidyNode tnod, int indent )
{
TidyNode child;
for ( child = tidyGetChild(tnod); child; child = tidyGetNext(child) )
{
ctmbstr name = tidyNodeGetName( child );
if ( name )
{
/* if it has a name, then it's an HTML tag ... */
TidyAttr attr;
printf( "%*.*s%s ", indent, indent, "<", name);
/* walk the attribute list */
for ( attr=tidyAttrFirst(child); attr; attr=tidyAttrNext(attr) ) {
printf(tidyAttrName(attr));
tidyAttrValue(attr)?printf("=\"%s\" ",
tidyAttrValue(attr)):printf(" ");
}
printf( ">\n");
}
else {
/* if it doesn't have a name, then it's probably text, cdata, etc... */
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetText(doc, child, &buf);
printf("%*.*s\n", indent, indent, buf.bp?(char *)buf.bp:"");
tidyBufFree(&buf);
}
dumpNode( doc, child, indent + 4 ); /* recursive */
}
}
QString UCHome_Main_SiteConst::find_sys_notice_by_type(TidyDoc doc, char *type_class)
{
QString note;
char *nid = type_class;
TidyNode node = NULL;
TidyNode node2 = NULL;
TidyBuffer tbuf;
if(nid == NULL) {
q_debug()<<"";
return note;
}
node = this->searchNode(doc, NULL, nid, TidyTag_DIV);
if(node != NULL) {
tidyBufInit(&tbuf);
tidyNodeGetText(doc, node, &tbuf);
note = this->u8codec->toUnicode(QByteArray((char*)tbuf.bp));
tidyBufFree(&tbuf);
q_debug()<<note;
}else{
q_debug()<<"Warning: "<<nid<<" not found";
}
return note;
}
bool CCFHtmlTidy::TidyMain(const char* pSourceIn, const char* pOptions, std::string &strOut, std::string &strErr)
{
TidyBuffer output;
TidyBuffer errbuf;
int rc = -1;
Bool ok = yes;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
tidyBufInit(&output);
tidyBufInit(&errbuf);
TidyOptionsSet(tidyDocToImpl(tdoc), pOptions);
if (ok)
rc = tidySetErrorBuffer(tdoc, &errbuf); // Capture diagnostics
if (rc >= 0)
rc = tidyParseString(tdoc, pSourceIn); // Parse the input
if (rc >= 0)
rc = tidyCleanAndRepair(tdoc); // Tidy it up!
if (rc >= 0)
rc = tidyRunDiagnostics(tdoc); // Kvetch
//if ( rc > 1 ) // If error, force output.
// rc = ( tidyOptSetBool(tdoc, TidyForceOutput, yes) ? rc : -1 );
if (rc >= 0)
rc = tidySaveBuffer(tdoc, &output); // Pretty Print
if (rc >= 0)
{
if (output.bp)
{
strOut = reinterpret_cast<char const*>(output.bp);
}
}
strErr = reinterpret_cast<char const*>(errbuf.bp);
std::string strEmpty = "No warnings or errors were found.\n\n";
if (0 == strEmpty.compare(strErr))
{
strErr.clear();
}
tidyBufFree(&output);
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return true;
}
// formats the 'dom' tree into HTML output.
int lua_tidy_saveString(lua_State *L)
{
pTidy t = toTidy(L,1);
TidyBuffer output;
tidyBufInit(&output);
lua_pushnumber(L, tidySaveBuffer(t->tdoc, &output));
lua_pushlstring(L, (char*)output.bp, output.size);
tidyBufFree( &output );
return 1;
}
bool HTidyInterface::formatSource( const char* textIn, CString &strTidy, CString &strMsg )
{
TidyBuffer output;
TidyBuffer errbuf;
int rc = -1;
Bool ok = yes;
TidyDoc tdoc = tidyCreate(); // Initialize "document"
tidyBufInit(&output);
tidyBufInit(&errbuf);
InitTidyDefault(tdoc);
SetTidyConfig(tdoc);
if ( ok )
rc = tidySetErrorBuffer(tdoc, &errbuf); // Capture diagnostics
if ( rc >= 0 )
rc = tidyParseString(tdoc, textIn); // Parse the input
if ( rc >= 0 )
rc = tidyCleanAndRepair(tdoc); // Tidy it up!
if ( rc >= 0 )
rc = tidyRunDiagnostics(tdoc); // Kvetch
//if ( rc > 1 ) // If error, force output.
// rc = ( tidyOptSetBool(tdoc, TidyForceOutput, yes) ? rc : -1 );
if ( rc >= 0 )
rc = tidySaveBuffer(tdoc, &output); // Pretty Print
if ( rc >= 0 )
{
strTidy = reinterpret_cast< char const* >(output.bp);
}
strMsg = reinterpret_cast< char const* >(errbuf.bp);
CString strEmpty = _T("No warnings or errors were found.\r\n\r\n");
if (0 == strEmpty.Compare(strMsg))
{
strMsg.Empty();
}
tidyBufFree(&output);
tidyBufFree(&errbuf);
tidyRelease(tdoc);
return true;
}
void CCaHtmlParse::__GetAirPortCode(CStringA & straDCode, CStringA & straACode, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
CStringA straSha("上海虹桥");
CStringA straPvg("上海浦东");
CStringA straPek("北京首都");
CStringA straNay("北京南苑");
straDCode = "";
straACode = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
CStringA straKey("<br />");//<br />后有回车换行符
int iPos = straRet.Find(straKey);
CStringA straDCity = straRet.Mid(4, iPos-4);
if (-1 != straDCity.Find(straSha))
straDCode = "SHA";
else if (-1 != straDCity.Find(straPvg))
straDCode = "PVG";
else if (-1 != straDCity.Find(straPek))
straDCode = "PEK";
else if(-1 != straDCity.Find(straNay))
straDCode = "NAY";
else
{
}
//取到达机场，<br />后有回车换行符
straRet = straRet.Mid(iPos+straKey.GetLength());
straRet.Remove(0x0d);//去掉回车
straRet.Remove(0x0a);//去掉换行
int iEndPos = straRet.Find("</td>");
CStringA straACity = straRet.Left(iEndPos);
if (-1 != straACity.Find(straSha))
straACode = "SHA";
else if (-1 != straACity.Find(straPvg))
straACode = "PVG";
else if (-1 != straACity.Find(straPek))
straACode = "PEK";
else if(-1 != straACity.Find(straNay))
straACode = "NAY";
else
{
}
tidyBufFree(&text);
}
/* {{{ proto string tidy_get_output()
Return a string representing the parsed tidy markup */
static PHP_FUNCTION(tidy_get_output)
{
TidyBuffer output;
TIDY_FETCH_OBJECT;
tidyBufInit(&output);
tidySaveBuffer(obj->ptdoc->doc, &output);
FIX_BUFFER(&output);
RETVAL_STRINGL((char *) output.bp, output.size ? output.size-1 : 0);
tidyBufFree(&output);
}
static int php_tidy_output_handler(void **nothing, php_output_context *output_context)
{
int status = FAILURE;
TidyDoc doc;
TidyBuffer inbuf, outbuf, errbuf;
if (TG(clean_output) && (output_context->op & PHP_OUTPUT_HANDLER_START) && (output_context->op & PHP_OUTPUT_HANDLER_FINAL)) {
doc = tidyCreate();
tidyBufInit(&errbuf);
if (0 == tidySetErrorBuffer(doc, &errbuf)) {
tidyOptSetBool(doc, TidyForceOutput, yes);
tidyOptSetBool(doc, TidyMark, no);
if (ZEND_SIZE_T_UINT_OVFL(output_context->in.used)) {
php_error_docref(NULL, E_WARNING, "Input string is too long");
return status;
}
TIDY_SET_DEFAULT_CONFIG(doc);
tidyBufInit(&inbuf);
tidyBufAttach(&inbuf, (byte *) output_context->in.data, (uint)output_context->in.used);
if (0 <= tidyParseBuffer(doc, &inbuf) && 0 <= tidyCleanAndRepair(doc)) {
tidyBufInit(&outbuf);
tidySaveBuffer(doc, &outbuf);
FIX_BUFFER(&outbuf);
output_context->out.data = (char *) outbuf.bp;
output_context->out.used = outbuf.size ? outbuf.size-1 : 0;
output_context->out.free = 1;
status = SUCCESS;
}
}
tidyRelease(doc);
tidyBufFree(&errbuf);
}
return status;
}
int main(int argc, char **argv )
{
CURL *curl;
char curl_errbuf[CURL_ERROR_SIZE];
TidyDoc tdoc;
TidyBuffer docbuf = {0};
TidyBuffer tidy_errbuf = {0};
int err;
if ( argc == 2) {
curl = curl_easy_init();
curl_easy_setopt(curl, CURLOPT_URL, argv[1]);
curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errbuf);
curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 0L);
curl_easy_setopt(curl, CURLOPT_VERBOSE, 1L);
curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_cb);
tdoc = tidyCreate();
tidyOptSetBool(tdoc, TidyForceOutput, yes); /* try harder */
tidyOptSetInt(tdoc, TidyWrapLen, 4096);
tidySetErrorBuffer( tdoc, &tidy_errbuf );
tidyBufInit(&docbuf);
curl_easy_setopt(curl, CURLOPT_WRITEDATA, &docbuf);
err=curl_easy_perform(curl);
if ( !err ) {
err = tidyParseBuffer(tdoc, &docbuf); /* parse the input */
if ( err >= 0 ) {
err = tidyCleanAndRepair(tdoc); /* fix any problems */
if ( err >= 0 ) {
err = tidyRunDiagnostics(tdoc); /* load tidy error buffer */
if ( err >= 0 ) {
dumpNode( tdoc, tidyGetRoot(tdoc), 0 ); /* walk the tree */
fprintf(stderr, "%s\n", tidy_errbuf.bp); /* show errors */
}
}
}
}
else
fprintf(stderr, "%s\n", curl_errbuf);
/* clean-up */
curl_easy_cleanup(curl);
tidyBufFree(&docbuf);
tidyBufFree(&tidy_errbuf);
tidyRelease(tdoc);
return(err);
}
else
printf( "usage: %s <url>\n", argv[0] );
return(0);
}
// quick and dirty shortcut function.
int lua_tidy_easyClean ( lua_State *L )
{
TidyBuffer output;
TidyBuffer errbuf;
int rc;
pTidy t;
const char * input;
tidyBufInit(&output);
tidyBufInit(&errbuf);
rc = -1;
t = toTidy(L,1);
input = lua_tostring(L,2);
rc = tidySetErrorBuffer( t->tdoc, &errbuf );
if ( rc >= 0 )
rc = tidyParseString( t->tdoc, input );
if ( rc >= 0 )
rc = tidyCleanAndRepair( t->tdoc );
if ( rc >= 0 )
rc = tidyRunDiagnostics( t->tdoc );
if ( rc >= 0 )
rc = tidySaveBuffer( t->tdoc, &output );
lua_pushlstring(L, (char*)output.bp,output.size);
if ( rc != 0 )
lua_pushlstring(L, (char*)errbuf.bp,errbuf.size);
else
lua_pushnil(L);
lua_pushnumber(L, rc);
tidyBufFree( &output );
tidyBufFree( &errbuf );
return 3;
}
char *copyTidyNodeValue(TidyNode tnode, TidyDoc tdoc)
{
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc,tnode,&buf);
char *str = (char *)xmalloc(buf.size+1);
memcpy(str,buf.bp,buf.size);
str[buf.size] = '\0';
tidyBufFree(&buf);
return str;
}
FeedRecord* UCHome_Main_SiteConst::parse_note(void *state_data, TidyDoc doc, TidyNode node)
{
FeedRecord * rec = NULL;
QString note;
TidyBuffer tbuf;
tidyBufInit(&tbuf);
tidyNodeGetText(doc, node, &tbuf);
note = this->u8codec->toUnicode(QByteArray((char*)tbuf.bp));
rec = new FeedRecord();
rec->content = this->rewrite_relative_link(note);
md5CheckSum((char*)tbuf.bp, tbuf.size, rec->md5sum);
tidyBufFree(&tbuf);
//q_debug()<<"Orig:"<<note;
return rec;
}
void CCaHtmlParse::__GetFlightStartTime(CStringA & strFlightStartTime, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
strFlightStartTime = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
CStringA straKey("\">");
int iStartPos = straRet.Find(straKey);
int iEndPos = straRet.Find("</strong>");
iStartPos = iStartPos+straKey.GetLength();
strFlightStartTime = straRet.Mid(iStartPos, iEndPos-iStartPos);
tidyBufFree(&text);
}
FeedRecord* UCHome_Main_SiteConst::parse_friend(void *state_data, TidyDoc doc, TidyNode node)
{
FeedRecord * rec = NULL;
QString note;
TidyBuffer tbuf;
TidyNode node2 = NULL;
TidyNode node3 = NULL;
TidyNode node4 = NULL;
TidyNode node5 = NULL;
TidyAttr attr = NULL;
ctmbstr fuid = NULL;
ctmbstr fusername = NULL;
tidyBufInit(&tbuf);
tidyNodeGetText(doc, node, &tbuf);
note = this->u8codec->toUnicode(QByteArray((char*)tbuf.bp));
rec = new FeedRecord();
rec->content = this->rewrite_relative_link(note);
rec->content = "<table>" + rec->content + "</table>";
md5CheckSum((char*)tbuf.bp, tbuf.size, rec->md5sum);
tidyBufFree(&tbuf);
//q_debug()<<"Orig:"<<note;
node2 = tidyGetChild(node);
node3 = tidyGetChild(node2);
attr = tidyAttrGetById(node3, TidyAttr_VALUE);
fuid = tidyAttrValue(attr);
node4 = tidyGetNext(node2);
node3 = tidyGetChild(node4); // A
node5 = tidyGetChild(node3); //IMG
attr = tidyAttrGetById(node5, TidyAttr_ALT);
fusername = tidyAttrValue(attr);
rec->fuid = fuid;
rec->fusername = this->u8codec->toUnicode(QByteArray(fusername));
q_debug()<<"USER:"<<fuid<<rec->fusername;
md5CheckSum(fuid, strlen(fuid), rec->md5sum);//使用uid的md5值肯定不会出现冲突
return rec;
}
static zend_object *tidy_object_new(zend_class_entry *class_type, zend_object_handlers *handlers, tidy_obj_type objtype)
{
PHPTidyObj *intern;
intern = ecalloc(1, sizeof(PHPTidyObj) + zend_object_properties_size(class_type));
zend_object_std_init(&intern->std, class_type);
object_properties_init(&intern->std, class_type);
switch(objtype) {
case is_node:
break;
case is_doc:
intern->ptdoc = emalloc(sizeof(PHPTidyDoc));
intern->ptdoc->doc = tidyCreate();
intern->ptdoc->ref_count = 1;
intern->ptdoc->initialized = 0;
intern->ptdoc->errbuf = emalloc(sizeof(TidyBuffer));
tidyBufInit(intern->ptdoc->errbuf);
if (tidySetErrorBuffer(intern->ptdoc->doc, intern->ptdoc->errbuf) != 0) {
tidyBufFree(intern->ptdoc->errbuf);
efree(intern->ptdoc->errbuf);
tidyRelease(intern->ptdoc->doc);
efree(intern->ptdoc);
efree(intern);
php_error_docref(NULL, E_ERROR, "Could not set Tidy error buffer");
}
tidyOptSetBool(intern->ptdoc->doc, TidyForceOutput, yes);
tidyOptSetBool(intern->ptdoc->doc, TidyMark, no);
TIDY_SET_DEFAULT_CONFIG(intern->ptdoc->doc);
tidy_add_default_properties(intern, is_doc);
break;
}
intern->std.handlers = handlers;
return &intern->std;
}
void CCaHtmlParse::__GetSaleEndDate(CStringA & straEndDate, CStringA & straEndTime, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
straEndDate = "";
straEndTime = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
int iPos = straRet.Find("<br");
straRet = straRet.Mid(4, iPos-4);
straEndDate = straRet.Left(10);
straEndTime = straRet.Mid(11);
straEndTime.Trim();
tidyBufFree(&text);
}
FeedRecord* UCHome_Main_SiteConst::parse_feed(void *state_data, TidyDoc doc, TidyNode node)
{
FeedRecord *rec = NULL;
QString feed_text ;
QByteArray feed_bytes;
int nid = 0;
ctmbstr fid = NULL;
TidyAttr attr = NULL;
TidyBuffer tbuf = {0};
tidyBufInit(&tbuf);
nid = tidyNodeGetId(node);
this->get_node_text(doc, node, &tbuf);
feed_bytes = QByteArray((char*)tbuf.bp);
feed_text = this->u8codec->toUnicode(feed_bytes);
tidyBufFree(&tbuf);
rec = new FeedRecord();
rec->content = feed_text;
//<li class="type_1006780" id="feed_685697_li">
//这个id是唯一的，用这个做标识
attr = tidyAttrGetById(node, TidyAttr_ID);
fid = tidyAttrValue(attr);
//q_debug()<<attr<<fid;
if(fid == NULL) {
q_debug()<<"Warning: invalid fid"<<fid<<feed_text;
return NULL;
}else if(strstr(fid, " 1000 ") != NULL) {
feed_bytes = fid;
feed_bytes = feed_bytes.replace(" 1000 ", "");
fid = feed_bytes.data();
q_debug()<<"Warning: invalid fid"<<tidyAttrValue(attr);
md5CheckSum(fid, strlen(fid), rec->md5sum);
}else{
md5CheckSum(fid, strlen(fid), rec->md5sum);
}
return rec;
}
void nuiHTMLNode::SetFromNode(const void* _tdoc, const void* _tnod, nglTextEncoding encoding, bool ComputeStyle)
{
Clear();
TidyDoc tdoc = (TidyDoc)_tdoc;
TidyNode tnod = (TidyNode)_tnod;
mName = nglString(tidyNodeGetName(tnod), encoding);
mType = (NodeType)tidyNodeGetType(tnod);
mTagType = (TagType)tidyNodeGetId(tnod);
nglString text;
TidyBuffer buf;
tidyBufInit(&buf);
if (tidyNodeGetValue(tdoc, tnod, &buf))
{
mText.Import((const char*)buf.bp, (int32)buf.size, encoding);
//NGL_OUT(_T("<%s> %s\n"), mName.GetChars(), mText.GetChars());
}
tidyBufFree(&buf);
// Fill the attributes:
TidyAttr tattr;
for (tattr = tidyAttrFirst(tnod); tattr; tattr = tidyAttrNext(tattr))
{
nuiHTMLAttrib* pAttrib = new nuiHTMLAttrib(tattr, encoding);
mAttributes.push_back(pAttrib);
}
if (ComputeStyle)
{
mpStyle = new nuiCSSStyle(this);
nuiHTMLAttrib* pStyle = GetAttribute(nuiHTMLAttrib::eAttrib_STYLE);
if (pStyle)
{
AddStyleSheet(GetSourceURL(), pStyle->GetValue(), true);
}
}
}
void CCaHtmlParse::__GetFlightNoAndFlightStartDate(CStringA & strFlightNo, CStringA & strFlightStartDate, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
strFlightNo = "";
strFlightStartDate = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
strFlightStartDate = straRet.Mid(4, 10);
int iStartPos = straRet.Find('C');//匹配CA，ca，Ca，cA，<br />后会插入回车换行符，
if(-1 == iStartPos)//没找到大写C，匹配小写c
iStartPos = straRet.Find('c');
int iEndPos = straRet.Find("</");
strFlightNo = straRet.Mid(iStartPos, iEndPos-iStartPos);//国航航班号有3位的，还有4位的
strFlightNo = strFlightNo.MakeUpper();
strFlightNo = strFlightNo.Mid(2);//去掉CA
tidyBufFree(&text);
}
static int php_tidy_parse_string(PHPTidyObj *obj, char *string, uint len, char *enc)
{
TidyBuffer buf;
if(enc) {
if (tidySetCharEncoding(obj->ptdoc->doc, enc) < 0) {
php_error_docref(NULL, E_WARNING, "Could not set encoding '%s'", enc);
return FAILURE;
}
}
obj->ptdoc->initialized = 1;
tidyBufInit(&buf);
tidyBufAttach(&buf, (byte *) string, len);
if (tidyParseBuffer(obj->ptdoc->doc, &buf) < 0) {
php_error_docref(NULL, E_WARNING, "%s", obj->ptdoc->errbuf->bp);
return FAILURE;
}
tidy_doc_update_properties(obj);
return SUCCESS;
}
QString UCHome_Main_SiteConst::get_time_string(TidyDoc doc, TidyNode tnode)
{
QString time_string ;
TidyBuffer tbuf = {0};
tidyBufInit(&tbuf);
tidyNodeGetText(doc, tnode, &tbuf);
if(strstr((char*)tbuf.bp, "2008-") != NULL) {
QDate onlyDate = QDate::fromString(QString((char*)tbuf.bp).trimmed(), "yyyy-MM-dd");
QDateTime currDate = QDateTime::currentDateTime();
currDate.setDate(onlyDate);
//q_debug()<<currDate<<onlyDate<<((char*)tbuf.bp);
time_string = onlyDate.toString();
}else{
//must yesterday
QDateTime currDate = QDateTime::currentDateTime().addDays(-1);
time_string = currDate.date().toString();
}
tidyBufFree(&tbuf);
return time_string;
}
static int tidy_node_cast_handler(zval *in, zval *out, int type)
{
TidyBuffer buf;
PHPTidyObj *obj;
switch(type) {
case IS_LONG:
ZVAL_LONG(out, 0);
break;
case IS_DOUBLE:
ZVAL_DOUBLE(out, 0);
break;
case _IS_BOOL:
ZVAL_TRUE(out);
break;
case IS_STRING:
obj = Z_TIDY_P(in);
tidyBufInit(&buf);
if (obj->ptdoc) {
tidyNodeGetText(obj->ptdoc->doc, obj->node, &buf);
ZVAL_STRINGL(out, (char *) buf.bp, buf.size-1);
} else {
ZVAL_EMPTY_STRING(out);
}
tidyBufFree(&buf);
break;
default:
return FAILURE;
}
return SUCCESS;
}
QString UCHome_Main_SiteConst::find_sign_text(TidyDoc doc)
{
QString sign_text;
TidyNode node;
TidyNode node2;
TidyBuffer tbuf;
node = this->searchNode(doc, NULL, "state", TidyTag_DIV);
if(node != NULL) {
node2 = tidyGetChild(node);
Q_ASSERT(tidyNodeGetId(node2) == TidyTag_A);
tidyBufInit(&tbuf);
if(tidyNodeGetText(doc, tidyGetChild(node2), &tbuf)) {
sign_text = this->u8codec->toUnicode(QByteArray((char*)tbuf.bp));
sign_text = sign_text.trimmed();
q_debug()<<"Sign text:"<<sign_text;
}
tidyBufFree(&tbuf);
}else{
q_debug()<<"Warning: no state sign text found";
}
return sign_text;
}
static void tidy_add_default_properties(PHPTidyObj *obj, tidy_obj_type type)
{
TidyBuffer buf;
TidyAttr tempattr;
TidyNode tempnode;
zval attribute, children, temp;
PHPTidyObj *newobj;
switch(type) {
case is_node:
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
tidyBufInit(&buf);
tidyNodeGetText(obj->ptdoc->doc, obj->node, &buf);
ADD_PROPERTY_STRINGL(obj->std.properties, value, buf.bp, buf.size ? buf.size-1 : 0);
tidyBufFree(&buf);
ADD_PROPERTY_STRING(obj->std.properties, name, tidyNodeGetName(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, type, tidyNodeGetType(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, line, tidyNodeLine(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, column, tidyNodeColumn(obj->node));
ADD_PROPERTY_BOOL(obj->std.properties, proprietary, tidyNodeIsProp(obj->ptdoc->doc, obj->node));
switch(tidyNodeGetType(obj->node)) {
case TidyNode_Root:
case TidyNode_DocType:
case TidyNode_Text:
case TidyNode_Comment:
break;
default:
ADD_PROPERTY_LONG(obj->std.properties, id, tidyNodeGetId(obj->node));
}
tempattr = tidyAttrFirst(obj->node);
if (tempattr) {
char *name, *val;
array_init(&attribute);
do {
name = (char *)tidyAttrName(tempattr);
val = (char *)tidyAttrValue(tempattr);
if (name && val) {
add_assoc_string(&attribute, name, val);
}
} while((tempattr = tidyAttrNext(tempattr)));
} else {
ZVAL_NULL(&attribute);
}
zend_hash_str_update(obj->std.properties, "attribute", sizeof("attribute") - 1, &attribute);
tempnode = tidyGetChild(obj->node);
if (tempnode) {
array_init(&children);
do {
tidy_instanciate(tidy_ce_node, &temp);
newobj = Z_TIDY_P(&temp);
newobj->node = tempnode;
newobj->type = is_node;
newobj->ptdoc = obj->ptdoc;
newobj->ptdoc->ref_count++;
tidy_add_default_properties(newobj, is_node);
add_next_index_zval(&children, &temp);
} while((tempnode = tidyGetNext(tempnode)));
} else {
ZVAL_NULL(&children);
}
zend_hash_str_update(obj->std.properties, "child", sizeof("child") - 1, &children);
break;
case is_doc:
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
ADD_PROPERTY_NULL(obj->std.properties, errorBuffer);
ADD_PROPERTY_NULL(obj->std.properties, value);
break;
}
}
static void php_tidy_quick_repair(INTERNAL_FUNCTION_PARAMETERS, zend_bool is_file)
{
char *enc = NULL;
size_t enc_len = 0;
zend_bool use_include_path = 0;
TidyDoc doc;
TidyBuffer *errbuf;
zend_string *data, *arg1;
zval *config = NULL;
if (is_file) {
if (zend_parse_parameters(ZEND_NUM_ARGS(), "P|zsb", &arg1, &config, &enc, &enc_len, &use_include_path) == FAILURE) {
RETURN_FALSE;
}
if (!(data = php_tidy_file_to_mem(ZSTR_VAL(arg1), use_include_path))) {
RETURN_FALSE;
}
} else {
if (zend_parse_parameters(ZEND_NUM_ARGS(), "S|zsb", &arg1, &config, &enc, &enc_len, &use_include_path) == FAILURE) {
RETURN_FALSE;
}
data = arg1;
}
if (ZEND_SIZE_T_UINT_OVFL(ZSTR_LEN(data))) {
php_error_docref(NULL, E_WARNING, "Input string is too long");
RETURN_FALSE;
}
doc = tidyCreate();
errbuf = emalloc(sizeof(TidyBuffer));
tidyBufInit(errbuf);
if (tidySetErrorBuffer(doc, errbuf) != 0) {
tidyBufFree(errbuf);
efree(errbuf);
tidyRelease(doc);
php_error_docref(NULL, E_ERROR, "Could not set Tidy error buffer");
}
tidyOptSetBool(doc, TidyForceOutput, yes);
tidyOptSetBool(doc, TidyMark, no);
TIDY_SET_DEFAULT_CONFIG(doc);
if (config) {
TIDY_APPLY_CONFIG_ZVAL(doc, config);
}
if(enc_len) {
if (tidySetCharEncoding(doc, enc) < 0) {
php_error_docref(NULL, E_WARNING, "Could not set encoding '%s'", enc);
RETVAL_FALSE;
}
}
if (data) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyBufAttach(&buf, (byte *) ZSTR_VAL(data), (uint)ZSTR_LEN(data));
if (tidyParseBuffer(doc, &buf) < 0) {
php_error_docref(NULL, E_WARNING, "%s", errbuf->bp);
RETVAL_FALSE;
} else {
if (tidyCleanAndRepair(doc) >= 0) {
TidyBuffer output;
tidyBufInit(&output);
tidySaveBuffer (doc, &output);
FIX_BUFFER(&output);
RETVAL_STRINGL((char *) output.bp, output.size ? output.size-1 : 0);
tidyBufFree(&output);
} else {
RETVAL_FALSE;
}
}
}
if (is_file) {
zend_string_release(data);
}
tidyBufFree(errbuf);
efree(errbuf);
tidyRelease(doc);
}